Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
データ分析基盤の変遷とデータレイクの作り方
Search
Sponsored
·
Ship Features Fearlessly
Turn features on and off without deploys. Used by thousands of Ruby developers.
→
Ojima Hikaru
April 21, 2018
Technology
1.9k
2
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
データ分析基盤の変遷とデータレイクの作り方
Battle Conference U30 #2018
Ojima Hikaru
April 21, 2018
More Decks by Ojima Hikaru
See All by Ojima Hikaru
家族の思い出を形にする 〜 1秒動画の生成を支えるインフラアーキテクチャ
ojima_h
3
2.4k
Railsの限界を超えろ!「家族アルバム みてね」の画像・動画の大規模アップロードを支えるアーキテクチャの変遷
ojima_h
5
1.3k
Podのオートスケーリングに苦戦し続けている話
ojima_h
1
430
ディメンショナルモデリングのすすめ
ojima_h
8
4.9k
モンスターストライクを支えるデータ分析基盤と準リアルタイム集計
ojima_h
7
5.9k
Other Decks in Technology
See All in Technology
dbt と Snowflake と tag
kevinrobot34
1
250
パスキーでドライブする アカウント統合(OAuth/OIDC Numa (Immersion) Workshop 2026)
oidfj
PRO
0
270
Rust×eBPFでEDRっぽいものをつくる
sunlife3
2
970
IHV like なユースケースへのOpenID Connect 関連仕様の適用事例
optim
0
270
マイナンバーカード本人確認の実装比較(OAuth/OIDC Numa (Immersion) Workshop 2026) / 20260825 numa-12
oidfj
PRO
0
250
VLMで2.3万枚のPyCon JP写真を検索!
terapyon
1
500
紙 to デジタル
ichien178
0
120
Oracle MCP Servers Explained
thatjeffsmith
1
500
「面白い!」を信じ抜け。激動の時代を貫く、オンリーワン・エンジニアの条件
kizawa2020
0
230
Introduction to Bill One Development Engineer
sansan33
PRO
0
470
GopherCon @シアトル に行ってきました
logica0419
0
240
NANDでも描画したい!
nichica906
3
730
Featured
See All Featured
Keith and Marios Guide to Fast Websites
keithpitt
413
23k
Fireside Chat
paigeccino
42
4k
More Than Pixels: Becoming A User Experience Designer
marktimemedia
3
500
Agile Actions for Facilitating Distributed Teams - ADO2019
mkilby
0
260
Measuring & Analyzing Core Web Vitals
bluesmoon
9
970
Rebuilding a faster, lazier Slack
samanthasiow
85
9.6k
The Mindset for Success: Future Career Progression
greggifford
PRO
0
480
Navigating the moral maze — ethical principles for Al-driven product design
skipperchong
2
500
Product Roadmaps are Hard
iamctodd
55
12k
The untapped power of vector embeddings
frankvandijk
2
1.8k
Mobile First: as difficult as doing things right
swwweet
225
10k
Producing Creativity
orderedlist
PRO
348
40k
Transcript
L FG A
• S')1 0(6T • L>A9 XFLAG CDB=
!?NRK • GRD /%Q$7 • GRDO:>3GRD;<8H;C-,/ ACFM • P?/5#2(4&"Q 1+/GRDJPR • BIERN/ • @RIC. *6 / • GitHub: ojima-h 2
4 DAUKPI !
5
6 • • 2TB/day
30 → 1000
7 • 5
→ 100
− 8 S3
− 9 S3
− 10 Redshift
− 11
12 Data Lake Architecture
Data Lake " • -4,&$#!-4,+.' • -4,&% "%,(13*+)40&% !
(Schema on Read) • Data Lake -4,& DWH 24/$ $% 13
Data Lake 14 Hive Metastore
Hive Metastore 15
Hive " • Hadoop%(47-:.69!; • SQL ,*7&$S3 # HDFS !1:/
#1:/ & • ORC !3')83+:502& 16
Hive Metastore • S3/HDFS * "-SQL /1,&(.&0 (.&%)! •
,&(.& • * "- • * "-*#.+') • (.&%$.+ • 17
Hive Metastore • EMR ! Hive Metastore
! • • EMR 30 18
Hive Metastore • Hive Metastore MySQL
• Hive Metastore (HCatalog) server • EMR 5 19
Hive Metastore S3 20
Hive Metastore • ' • '"%
• 'ORC • '!&' ' !'#$$ 21
Hive Metastore • Hive Metastore S3 "
S3" !" 22
Hive Metastore * • "+$%- :>:>(*+ • 8C6*/,# •
3C;4' Hive DB / • Hive ).!% S3&*8C6/ • Hive &.( 8C6)-*@C@/ 23 3C;4 D=A49B<019?C2BBE 8C6579 8C6 Hive Database Table Partition S3 s3://BUCKET/warehouse/SERVICE.db/ s3://BUCKET/warehouse/SERVICE.db/TABLE/ s3://BUCKET/warehouse/SERVICE.db/TABLE/y=YYYY/m=MM/d=DD/
Hive Metastore • %)" &'&'%)" • &$#
! ( 24
Hive Metastore 1. Hive Metastore
25
Hive Metastore 1. Hive Metastore
2. 26
Hive Metastore 1. Hive Metastore
2. 3. Hive Metastore 27
Hive Metastore 1. Hive Metastore
2. 3. Hive Metastore 4. 28
Hive Metastore ! 1. ),(! $ Hive Metastore # 2.
),($'*, 3. Hive Metastore ! $ 4. ),($ &%+ $ "),($ 29
Hive Metastore 30
Hive Metastore • Hive Redshift "%!$%# • Redshift
COPY "%! csv+gzip • Hive "%! ORC • Redshift csv+gzip Hive ORC ⇒ Redshift Spectrum 31
Redshift Spectrum • Redshift S3(#$+ &%*" • ',)+
Hive Metastore ! Hive ',)+" 32 CREATE EXTERNAL SCHEMA schema_name FROM HIVE METASTORE DATABASE 'database_name’ URI 'hive_metastore_uri’;
Hive Metastore • Redshift Hive 33 INSERT
INTO ‘Redshift ’ SELECT … FROM ‘Hive ’ WHERE y=YYYY AND m=MM AND d=DD;
Hive Metastore • Redshift Spectrum
Hive Metastore • Spark SQL • Presto • Athena • Flink 34
Hive Metastore Hive Metastore S3 Hive,
Redshift Spectrum , Spark 35
36
($) • Hive Metastore '25103-$251.4/4& • Hive Metastore , $"
Data Lake , !$# 251&*251&%+$#! Hive Metastore , +$# Data Lake , "$#(!6 37
None